Cisco changed the datacenter game with UCS – the only server designed from top to bottom for virtualization. Despite widespread skepticism that the networking giant knew nothing about servers and would fail miserably, in less than five years UCS became the number one blade seller in the Americas.
In our new compressed disruptive-cycle world, Cisco itself has now fallen far behind when it comes to optimally hosting a virtualized datacenter. The company is, however, scurrying to catch up. Within the past few weeks, it’s been revealed that Cisco both invested in hyper-converged startup, Stratoscale, and also blessed Maxta as the first, and so far only, certified hyper-converged solution to run on Cisco UCS.
How UCS Thumped the Server Leaders
I’ve been a vocal fan of UCS from the beginning. In late 2009, when “over 100 companies” were using UCS, I wrote a blog post comparing UCS vs. the HP Matrix. While competitors scoffed at UCS as a “one-size-fits-all product”, I maintained that it would revolutionize datacenter virtualization.
The dominant server manufacturers of the day were perfectly happy with the status quo. But Cisco realized that virtualization would become the datacenter standard and that a new type of server was required. Cisco initially approached IBM and HP to jointly develop a product, but both companies declined. So Cisco instead funded VMware cofounder, Ed Bugnion, and a team of engineers to spend three years building UCS.
UCS helped mitigate virtualization challenges with capabilities such as FCoE (Fibre Channel over Ethernet), hypervisor bypass, extended memory, services profiles and a GUI that can help the server, storage and network teams collaborate more effectively.
But UCS’s Achilles heel is that it really only addresses a small part of the virtualized datacenter issues – the compute. By far the majority of the pain in the modern datacenter has to do with storage. Not surprisingly, four storage manufacturers, EMC/VCE, NetApp, Hitachi and Nimble, all incorporate UCS as an integral component of their so-called “converged infrastructure” solutions.
Channel partners across the globe, such as the one I worked for, understood that as customers increasingly virtualized their datacenters, they would want the enterprise capabilities and features that UCS offered. These partners worked with Cisco to make UCS the number two blade seller in the world.
Descending into Irrelevance
Ah, but all things must change – especially in a software-defined world. While Cisco was promoting the superiority of custom-designed ASICs, Nutanix was bringing the advantages of commodity-driven web-scale architecture to the enterprise. The impressive innovations that Cisco unveiled over five years ago are now not just obsolete, but superfluous.
- Fibre Channel over Ethernet (FCoE): Unlike the converged infrastructure offerings built around UCS, FCoE is an example of true convergence of the network stack – melding fibre channel and IP Ethernet networks. But today, Web-scale eliminates the requirement for SANs and switching fabrics entirely.
- UCS Manager GUI: Lets storage and server teams collaborate more effectively together. Not so useful when separate storage administrators are no longer necessary.
- Custom ASICs: Cisco boasts 12% increased performance from proprietary hardware. Nice but inconsequential when Moore’s Law doubles performance every 18 months anyway. Nutanix utilizes commodity hardware, but increases performance nonetheless with regular software updates that improve hardware effectiveness.
- Services profiles and templates: These were great in the day for relatively fast provisioning of ESX hosts. Nutanix Foundation is much faster and doesn’t require zone masking or manual hypervisor installs.
- Integrating the Cisco Nexus switch: Making the network the management center was key to Cisco gaining traction with its network administrator constituency. But Web-scale eliminates the requirement for complex, intelligent and expensive converged network switches.
The leading converged infrastructure manufacturer, VCE, proudly advertises that it only takes 45 days to order and put a Vblock into production – 5 X faster than with conventional servers and storage. In contrast, Nutanix can be ordered, received, installed and in production in around five days.
Upgrading VMware vSphere requires a corresponding upgrade to the entire Vblock – a process that can easily require a team of consultants several days to accomplish. And even then there are risks involved. A former Vblock customer that recently migrated to Nutanix was still running three versions back of vSphere because they didn’t want to have to deal with the associated Vblock upgrade.
Contrast all of this time, expense and risk with doing a vSphere (or Hyper-V or KVM) upgrade on Nutanix. The process is literally just a single click. No cost, no downtime and no risk.
Lesson for Nutanix
UCS and Nutanix both target the same customers – virtualized enterprise environments. I’ve heard from multiple partners that despite our relatively tiny size, Cisco has declared Nutanix to be its number one competitor. Not HP. Not VMware. Nutanix. Cisco’s announcements around Maxta and Stratoscale reflect its determination to, albeit belatedly, get into the game.
Cisco is one of the most successful and well-run companies of all time. While known for its innovations in areas such as routing, switching, VoIP and collaboration – perhaps nothing has been as impressive as Cisco’s accomplishment in the datacenter. Cisco upended all of the existing dominant server players by developing UCS to fulfill the computing requirements of the virtualized datacenter.
The lesson here for Nutanix is that if Cisco can fall into complacency, anyone can. We’ve got to keep our heads down, be humble, stay hungry and keep innovating – even if we have to eventually disrupt our own technologies.
Lesson for Channel Partners
Cisco, VMware, Nutanix, Dell and HP, in addition to the other EVO:Rail partners and lots of startups, validate that hyper-converged infrastructure/web-scale is the future of the virtualized datacenter. There’s a $50 billion + annual server and storage market out there just begging to be disrupted by those channel partners with both the vision and the desire to execute.
Thanks to @vmmike130, @langonej, @evolvingneurons and to @richardarsenian for input.